Study On Algorithms Of 3-D Reconstruction From Serial Slice-images Based On Fractal

With the development of some irrigation works and rock projects, it is important to study the rock fractures. 3-Dimension (3-D) reconstruction of rock fractures surface by computer can show the interior architecture of rock. Based on the background, the 3-D visualization was studied for serial slice-images data in the paper. The process mainly includes some steps: image segmentation, noise filtering, lines fitting, detection of corner points, 3-D interpolation. Base on fractal theory, the paper paid attention to image segment by fractal algorithm and 3-D fractal interpolation.Image segmentation plays an important role in image processing. The stability of some traditional edge segment algorithms for rock image which has lower textural feature is worse. It is why fractal method is used. Fractal dimension is a statistic value, which can distinguish between the smooth and rough area of image. So the key difficulty of segmentation is fractal dimension. Many general fractal dimension algorithms of image were discussed. Base on the Differential Box Counting (DBC), the Seed Pixel Differential Box Counting (SPDBC) method was proposed. It is better to use the method for little image area or low resolution image. The image segmentation experiment shows that the method is effective and reliable.The distance between the slice-images is larger than general, so the sample data is lacking in the direction of vertical slice-images. But the advantage of fractal interpolation can get a large mount of data set from a small data set and make the new data set rough. A novel method Initial Domain Expanding Fractal Interpolation (IDEFI) based on Iterated Function System (IFS) is proposed for constructing fractal interpolation surfaces through points sampled on rectangular grids. The proposed algorithm guarantees the continuity of the boundary and cancels the restriction of symmetry on the initial data set. Compared with Re-ordering Mapping Fractal Interpolation (RMFI), IDEFI is swifter to iterate, easier to map. The result surface of IDEFI inherits the main features from the original data set as a whole and keeps the self-similar trait of fractal in part.The process of line fitting and detection of corner points transforms image data into 3-D graphic data. Another new algorithm based on fractal interpolation is presented for reconstructing 3-D surfaces. Unlike the existing general algorithms, which are appropriate only for dataset sampled on rectangular grids, the proposed algorithm can deal with the initial points defined on irregular quadrangular grids, which is shaped by serial cross-sectional data. The experimental result shows it is practical for 3-D reconstruction.